The purpose of this little exercise isn’t to decide whether or not it people should go vote.

The purpose to examine why people with certain preferences will choose to vote or not to vote in an election between two candidates based on payoffs and internal and external incentives. To begin, let’s assume that in voting, you have a utility function dependent on the following perceptions:
U = α + βλ − ϕ

α is a function of “patriotism” or “good citizenship,” generally measuring how much you value voting for the sake of voting. If you believe voting to be your civil duty, you’re more likely to generate some utility simply by going to the voting booth and pushing the button. This is also a function of internal and social pressures; if you obtain some utility from wearing an “I voted” sticker, then your α is higher to reflect that. You may also face pressures from external sources that work in the opposite direction: people around you that cause you disutility if you do not vote (for example, an angry wife that wishes you weren’t wasting your time on this analysis).

λ is a measure of the impact of your vote. For simplicity, I will assume that you care only about the degree that your vote has an impact and has no bearing on whether the public’s vote is positive or negative with respect to your own vote. To measure impact, I will use

λ=

[outcome if you do vote] − [outcome if you don't vote] [outcome if you don't vote]

β is the rate that you value your impact. If you care highly about your impact, your β will be high, and if you care almost nothing about your impact, your β will be close to 0. The coefficient is a function of candidate preference, your alignment on issues, and how much you expect the candidate to change impact your life (for the better or otherwise). ϕ is simply the cost of voting. It is the value of the time you spend traveling to the polling place and waiting in line, etc. This can also measure the extra time it costs and unregistered voter to get registered, assuming they can legally do so immediately before voting. As we will eventually compare our expected utility in voting with the expected utility in doing something else, we will not include opportunity cost in this variable.

For the purposes of evaluating a final decision, I’ll assume that the payoffs to voting and not voting are as follows: s0 v+0 0 s1 v + s1 s1 s2 v + s2 s2 s3 v + s3 s3 … … … sn v + sn sn

Vote Don’t vote • •

v = your vote for your candidate, and si = the absolute value of the difference between votes for your candidate and votes against your candidate, not counting your own vote, beginning with s0 = 0 other voters. This is just a measure of whether the election in question is a landslide or a nail-biter. Each si has a probability pi = the probability that a certain number of people will vote and that the difference between their votes is si. Thus, pisi gives a probability distribution of the likelihood of varying degrees of victory for each candidate.

For the purposes of this paper (and for the purposes of pretty much everything in nonbehavioral economics), I will assume that people are rational and that they maximize their expected utility; that is, they’ll do what they expect gives them the most satisfaction. Then E(U ) = E(α + βλ − ϕ ) = α + β E(λ ) − ϕ . That is, we know inherently (or subconsciously, at least) our own personal values for α, β and ϕ. But we must calculate the expected value of our impacts based on our perception of the probabilities given above. We will have
E(vote) = p0 (v) + p1 (v + s1 ) + p2 (v + s2 ) + K + pi (v + si ) = p0 v + p1v + p1s1 + p2 v + p3 s3 + K + pn v + pn sn = v( p0 + p1 + K + pn ) + p1s1 + p2 s2 + K + pn sn
n

= v + ∑ pi si
i =1

and
E(don 't) = p0 (0) + p1s1 + p2 s2 + K + pn sn
n

= ∑ pi si
i =1

Now we can estimate the expected impact of our vote:

n

n

(v + ∑ pi si ) − ∑ pi si E(λ) =
i =1 n i =1

=

v
n

∑ps
i =1

i i

∑ps
i =1

i i

Then our expected utility is given by
E(U ) = α + β ⋅[ v
n

]−ϕ
i i

∑ps
i =1

In deciding whether or not to vote, we have to compare our expected utility in voting with an expected utility of the alternative. Let’s call the alternative E(U’) and assume that it is higher than any other alternative. This will serve as our opportunity cost for voting. We know that if E(U) > E(U’), then someone rational will vote, and vice-versa:
Vote if α + β v
n

− ϕ > E(U ')
i i

∑ps
i =1

Don't vote if α + β

v
n

− ϕ < E(U ')
i i

∑ps
i =1

Now, we can measure the effects of our parameters on whether or not a rational person will vote:

As α increases, so does the expected utility of voting against the alternative. This illustrates how “get out the vote” and “vote or die” and “your vote counts” campaigns try to target voters’ expected utility for voting. The effectiveness of such campaigns vary, and from the six minutes’ research I just did, mostly impact independent voters. For many people, voting may be their only opportunity to explicitly participate in governmental proceedings. In that case, I’m sure that perception has a large impact on α and therefore the likelihood that those people will vote.

The expected impact of your vote depends mostly on pisi, the probability that the outcome excluding your vote is small or large. As the probability that si is small
n

increases, the weighted value of

∑ps
i =1

i i

decreases, increasing the expected utility.

This variable says that as your perception of the outcome excluding your vote is important to your expected impact in the election. If you believe the likelihood is high that the election will be a landslide in either direction, you are unlikely to vote. You are more likely to vote if you believe that you could cast (one of) the deciding vote(s) in the election.

β’s impact really measure’s how much you value your impact. If a voter cares only for her impact in a specific election and doesn’t put stock into civic duty or participation in this process, then she may have a high β and a low α. As β increases, she is more likely to derive utility from voting. But the effect could be low or negligible if she expects her impact to be very small. ϕ represents the cost of voting. It could easily be included in the figure E(U’) supposing the activity in question doesn’t include walking to the polling place or standing in line.
This is the figure some campaigns seek to diminish by combining registration and early voting, or by registering people on the sidewalk.

What does this model say that we don’t already know? Well, nothing. It only points out that voters derive utility from voting based on certain characteristics that they have developed or that have developed around them. Then should a rational person vote? My answer would be that a rational person should try to decide which parameter is important to them, and make a decision based on the answer.